Intelligent Autonomous Agents Ralf Möller, Rainer Marrone Hamburg University of Technology
Lab class Tutor: Rainer Marrone Time: Monday 12:15-13:00 Locaton: SBS93 A0.13.1/2 w Starting in Week 3
Literature Chapters 2, 6, 13, 15-17 http://aima.cs.berkeley.edu
Literature
Literature
Java Agent Development Framework http://jade.tilab.com/ 6
What is an Agent? An agent is anything that can be viewed as perceiving its environment through sensors and acting upon that environment through actuators Human agent: eyes, ears, and other organs for sensors; hands, legs, mouth, and other body parts for actuators Robotic agent: cameras and infrared range finders for sensors; various motors for actuators
Agents and environments The agent function maps from percept histories to actions: [f: P* à A] The agent program runs on the physical architecture to produce f Agent = architecture + program
Vacuum-Cleaner World Percepts: location and contents, e.g., [A,Dirty] Actions: Left, Right, Suck, NoOp
A Vacuum-Cleaner Agent
Capabilities An agent is capable of flexible action in some environment By flexible, we mean: w reactive w pro-active w social 11
Reactivity If a program s environment is guaranteed to be fixed, the program need never worry about its own success or failure program just executes blindly w Example of fixed environment: compiler The real world is not like that: things change, information is incomplete. Many (most?) interesting environments are dynamic A reactive system is one that maintains an ongoing interaction with its environment, and responds to changes that occur in it (in time for the response to be useful) 12
Proactiveness Reacting to an environment is easy (e.g., stimulus response rules) But we generally want agents to do things for us Hence goal directed behavior Pro-activeness = generating and attempting to achieve goals w Not driven solely by events w Taking the initiative w Recognizing opportunities 13
Balancing Reactive and Goal- Oriented Behavior We want our agents to be reactive, responding to changing conditions in an appropriate (timely) fashion We want our agents to systematically work towards long-term goals These two considerations can be at odds with one another Designing an agent that can balance the two remains an open research problem 14
Social Ability The real world is a multi-agent environment: we cannot go around attempting to achieve goals without taking others into account Some goals can only be achieved with the cooperation of others Social ability in agents is the ability to interact with other agents (and possibly humans) via some kind of agentcommunication language with the goal to let other agents to make commitments 15
Rational Agents An agent should strive to "do the right thing", based on what it can perceive and the actions it can perform. The right action is the one that will cause the agent to be most successful Success to be measured w.r.t. an agent-local perspective Performance measure: An objective criterion for success of an agent's behavior E.g., performance measure of a vacuum-cleaner agent could be amount of dirt cleaned up, amount of time taken, amount of electricity consumed, amount of noise generated, etc. 16
Rational Agents Rational Agent: For each possible percept sequence, a rational agent w should select an action w that is expected to maximize its local performance measure, w given the evidence provided by the percept sequence and w whatever built-in knowledge the agent has. Rational = Intelligent 17
Autonomous Agents Rationality is distinct from omniscience (allknowing with infinite knowledge) Agents can perform actions in order to modify future percepts so as to obtain useful information (information gathering, exploration) An agent is autonomous if its behavior is determined by its own experience (with ability to learn and adapt) 18
Other Properties Mobility: the ability of an agent to move around an electronic network, real environment (Veracity: an agent will not knowingly communicate false information) Benevolence: agents do not have conflicting goals, and every agent will therefore always try to do what is asked Learning/adaption: agents improve performance over time 19
Agents and Objects Are agents just objects by another name? Object: w Encapsulates some state w Interacts synchronously with other objects communicates via message passing call methods / generic functions w Has methods, corresponding to operations that may be performed on this state 20
Agents and Objects Main differences: w Agents are autonomous: agents embody stronger notion of autonomy than objects, and in particular, they decide for themselves whether or not to perform an action on request from another agent w Agents are capable of flexible (reactive, pro-active, social) behavior, and the standard object model has nothing to say about such types of behavior w Agents are inherently multi-threaded, in that each agent is assumed to have at least one thread of active control 21
Main Features Performance measure Environment Actuators Sensors Must first specify the setting for intelligent agent design
PEAS Consider, e.g., the task of designing an automated taxi driver: w Performance measure: Safe, fast, legal, comfortable trip, maximize profits w Environment: Roads, other traffic, pedestrians, customers w Actuators: Steering wheel, accelerator, brake, signal, horn w Sensors: Cameras, sonar, speedometer, GPS, odometer, engine sensors
PEAS Agent: Medical diagnosis system w Performance measure: Healthy patient, minimize costs, lawsuits w Environment: Patient, hospital, staff w Actuators: Screen display (questions, tests, diagnoses, treatments, referrals) w Sensors: Keyboard (entry of symptoms, findings, patient's answers)
PEAS Agent: Part-picking robot w Performance measure: Percentage of parts in correct bins relative to number of parts on the ground w Environment: Area with parts, bins w Actuators: Jointed arm and hand w Sensors: Camera, joint angle sensors
PEAS Agent: Interactive English tutor w Performance measure: Maximize student's score on test w Environment: Set of students w Actuators: Screen display (exercises, suggestions, corrections) w Sensors: Keyboard
Environment Types Fully observable (vs. partially observable): An agent's sensors give it access to the complete state of the environment at each point in time. Deterministic (vs. stochastic): The next state of the environment is completely determined by the current state and the action executed by the agent. (If the environment is deterministic except for the actions of other agents, then the environment is strategic) Episodic (vs. sequential): The agent's experience is divided into atomic "episodes" (each episode consists of the agent perceiving and then performing a single action), and the choice of action in each episode depends only on the episode itself.
Environment Types Static (vs. dynamic): The environment is unchanged while an agent is deliberating. (The environment is semidynamic if the environment itself does not change with the passage of time but the agent's performance score does) Discrete (vs. continuous): A limited number of distinct, clearly defined percepts and actions. Single agent (vs. multiagent): An agent operating by itself in an environment.
Environment Types Chess with Chess without Taxi driving a clock a clock Fully observable Yes Yes No Deterministic Strategic Strategic No Episodic No No No Static Semi Yes No Discrete Yes Yes No Single agent No No No The environment type largely determines the agent design The real world is (of course) partially observable, stochastic, sequential, dynamic, continuous, multi-agent
Environment Types Chess with Chess without Taxi driving a clock a clock Fully observable Yes Yes No Deterministic Strategic Strategic No Episodic No No No Static Semi Yes No Discrete Yes Yes No Single agent No No No The environment type largely determines the agent design The real world is (of course) partially observable, stochastic, sequential, dynamic, continuous, multi-agent
Environment Types Chess with Chess without Taxi driving a clock a clock Fully observable Yes Yes No Deterministic Strategic Strategic No Episodic No No No Static Semi Yes No Discrete Yes Yes No Single agent No No No The environment type largely determines the agent design The real world is (of course) partially observable, stochastic, sequential, dynamic, continuous, multi-agent
Environment Types Chess with Chess without Taxi driving a clock a clock Fully observable Yes Yes No Deterministic Strategic Strategic No Episodic No No No Static Semi Yes No Discrete Yes Yes No Single agent No No No The environment type largely determines the agent design The real world is (of course) partially observable, stochastic, sequential, dynamic, continuous, multi-agent
Agent Functions and Programs An agent is completely specified by the agent function mapping percept sequences to actions One agent function (or a small equivalence class) is rational Aim: find a way to implement the rational agent function concisely
Table-Lookup Agent Drawbacks: w Huge table w Take a long time to build the table w No autonomy w Even with learning, need a long time to learn the table entries
Agent Types Four basic types in order of increasing generality: w Simple reflex agents w Model-based reflex agents w Goal-based agents w Utility-based agents
Simple Reflex Agents
Model-Based Reflex Agents
Goal-Based Agents
Utility-Based Agents
Learning Agents